Author: Chris Han

How to track Google AI Mode without rankings or click data: a prompt-sampling method, sample-size math from 88,036 sampled answers, tool options, and a report template that survives scrutiny.

A content strategy for AI Mode lives or dies on facet coverage. We tracked 19,412 AI Mode citations for eight weeks to find when one deep hub beats five narrow pages — and the rule for splitting or merging.

Why AI engines give different answers about your brand: 3,600 tracked answers show retrieval causes 54% of divergence, model behavior 46%. See the four-layer breakdown and what fixes each.

Not all eight AI engines earn a dashboard slot. Data from 1.9M tracked answers shows which AI engines to track daily, weekly, and quarterly — plus the scoring formula to build your own list.

Claude AI brand recommendations name 48% fewer brands and hedge twice as often as ChatGPT. See the 11,520-answer panel data and what to fix first.

Cross-engine AI visibility consensus turns engine agreement into root-cause evidence. What 1,204 tracked step changes across 8 AI engines reveal about causes.

Which sources cited across AI engines overlap? We tracked 607,318 citations on 8 platforms: just 1.9% of URLs win six or more. See the five shared traits.

What actually improves AI visibility? A 630,000-run null test found 56% of reported AEO wins sit inside normal noise. Learn the evidence bar that proves cause.

How long for AI to update answers after an edit? A 312-edit panel with daily sampling gives per-engine medians, false-negative rates by check date, and the exact day a null result becomes real evidence.

AI visibility sample size, settled with real numbers: 23,040 tracked answers show why 200 prompts × 3 runs beats 20 × 30. Get the math and copy the design.